Q: What types of data are processed by the AI?
A: TalentAdore’s AI processes a range of data necessary for enhancing recruitment processes, including:
- Personally identifiable information (PII), such as candidate names and contact details.
- CVs and cover letters submitted by applicants.
- Job descriptions and related textual content provided by employers.
- Structured feedback and evaluation data from recruitment teams.
TalentAdore ensures that all data is handled securely and processed in compliance with GDPR and relevant data protection regulations.
Q: Where is data processed and stored?
A: All data is processed and stored within the EU/EEA zone to ensure compliance with GDPR and other relevant data protection regulations.
Q: How is customer data protected by TalentAdore when using LLM-powered features?
A:
- All customer data is stored and processed on TalentAdore’s own servers located within the EU/EEA zone.
- Data in transit and at rest is always encrypted using industry-standard encryption protocols.
- No data is sent to LLMs outside the EU where customer data is involved.
- AWS Bedrock acts as the infrastructure layer and enforces strict data security practices. Any data passed through Bedrock remains within the customer's AWS environment and is not stored or used by Anthropic or AWS to train models.
Q: How does the usage of LLM-powered features impact compliance with GDPR?
A: TalentAdore’s LLM powered features are fully compliant with GDPR. All data processing is conducted within the EU, ensuring compliance with regulations on data protection.
- No Profiling: TalentAdore’s AI does not engage in profiling or make decisions that produce legal effects or similarly significant impacts on individuals. Instead, the AI assists recruiters by generating content (e.g., summaries & feedbacks) based on human-provided inputs.
- No Automated Decision-Making: There is no fully automated decision-making process involved in the use of LLM-powered features. All AI-generated content is designed to assist and support recruiters, who remain in control of the final decisions and communications.
- Data Minimization and Purpose Limitation: Only the data necessary for generating specific outputs is processed, and the data is not stored or used beyond the intended purpose.
- User Control: Organizations have full control over the use of LLM- powered features. These features can be enabled or disabled at the company level, ensuring flexibility and alignment with organizational data policies.
Q: What types of data are used to train the AI? Is any customer-sensitive data involved?
A: TalentAdore does not use customer data to train AI models. Our LLM-powered features leverage pre-trained models and we do not store or feed customer-sensitive data back into these systems for training purposes.
Q: What kind of LLM (Large Language Model)/AI technology does TalentAdore use?
A: TalentAdore employs a hybrid architecture for AI services, primarily developed in-house. This ensures flexibility, as we are not tied to any single LLM provider. Currently, TalentAdore utilizes GPT technologies from Anthropic models hosted on AWS Bedrock for generating content.
Q: What is the basic structure of the data flow?
A:
- Data Collection: TalentAdore collects application data (e.g., CVs, cover letters, job descriptions) in textual form through the TalentAdore platform.
- Data Analysis: Contextual information is processed and prompts are formulated. These prompts are then sent to AI models (Those hosted on AWS Bedrock using Anthropic GPT technologies).
- Output Generation: The AI models analyze the inputs and generate responses, such as candidate communication templates or job descriptions.
- User Presentation: The results are presented to the user in real time or via batch processing in the background (e.g., summaries or pre-drafted communications).
Q: How can I trust the quality of the content generated by AI, especially for candidate communication?
A: TalentAdore has implemented a rules-based architecture to guide AI output, ensuring that text generation adheres to predefined instructions and parameters. Before final output, the system applies additional quality checks and validations to maintain accuracy, professionalism, and consistency.
Comments
0 comments
Please sign in to leave a comment.